Best Keyboards Ever: A Deep Dive (Documentary)

[esr]

Attached is a draft script for a YouTube documentary focused on buckling-spring keyboards. I'm submitting it here for comments and corrections.

deepdive.doc

(26.23 KiB) Downloaded 197 times

Readthrough time is 26 minutes. This is normal for long-form YouTube stuff, but we can't realistically make it much longer. So please try not to insist that we add obsessive amounts of detail.

[sharktastica]

Just something quick I've noticed. I know in general it makes sense for soft touch = rubber dome, but in IBM context, Soft Touch was what they called factory-greased buckling springs for P/N 8184692. IBM called their domes Quiet Touch, instead

I'll have a proper read tomorrow - looking forward to seeing this video.

[XMIT]

.doc

Please use one of the many open file formats. PDF, text file, .odf, or just pasting inline with a "spoiler" tag come to mind.

[depletedvespene]

Okay, so...

• Honorable mention to ((m)ass)drop?
• The entire thing is about buckling springs; the presence of a beamspring at the beginning is justified, but the Omnikey is barely mentioned as a side-note. Talk about it some more OR take it out the starting act and go full-on buckling spring fanboy (*).
• Another side note is present, deriding Cherry, but no mention is made of other keyboard switches that are well respected, even by buckling spring fanboys (*), like Space Invaders (which happen to be very much in the news lately).
• The draft's focus is clearly on the switches, but not enough is made of the actual form factor and physical layout of keyboards — the Enhanced layout (which was introduced on the Model M keyboards) is what makes many people actually prefer a Model M unit over an F AT or an F XT, despite the latter two being, well, F (better switches and NKRO). Heck, beamsprings have even more variation in this regard; showing a few of them should drive home the point of an evolving aspect of keyboards at the time.
• "Original Model F" refers to, we presume here, an F XT keyboard. Notably absent is an F AT (which should be used to point the evolution of the physical layouts until the Enhanced layout was established).
• Not enough of a separation is made between the evolution of terminal keyboards and PC keyboards (the latter spawned from the former), and why the heavy differing treatment of each group is made. DO mention that until Soarer's Converter came along, terminal keyboards (full-size, M122 and F122) were very much unwanted, good only as part donors if at all.

(*) Note that I myself am a buckling spring fanboy.

[esr]

.doc

Please use one of the many open file formats. PDF, text file, .odf, or just pasting inline with a "spoiler" tag come to mind.

It's plain text. I gave it a .doc extension only to pacify the uploader - it wouldn't take .adoc or .txt.

[xxhellfirexx]

IBM beamsprings have been made even in the mid-80's. See some IBM 3727 keyboards with the ^ symbol above the 6 instead of ¬. They switched over to buckling springs because it was too tall for the DIN standard. The height from the table to the top of the home row key must not be taller than 30 mm. They did everything they could to reduce the height, even switching to thin cork pads.

Buckling springs also has less parts to make and store. For doubleshot keycaps, you need molds for every key in every row in every language. And you need more molds for tripleshot caps. For buckling springs, due to the curved backplate and dye-sub legends, only 1 keycap needs to be made. The dyesub lettering are on sheets.

Maybe you could elaborate on the beamspring vs model F vs model M key feel and sound. Beamsprings have the lightest key feel of the three (~55g). It feels crisp, delicate, slightly tactile, effortless, and has a peculiar metallic sound.

Model Fs are slightly heavier (~60g). They are more tactile, effortless, and has a distinctive sharp buckling sound. However, some feel they feel and sound less refined. I don't really like the weird vibration sensation on my fingertips after a key is activated.

Model Ms are even heavier (~70g) so for some people, it starts to get tiring. They feel less tactile than model Fs and the buckling sound is more hollow and tinny and to me sounds wrong.

However, there is one thing that is the same for all these keyboards. They have a pinging sound that varies between boards even when they are made in the same factory in the same year. The way they are stored and the amount of use makes some more pingy than others and that gives them character. I can close my eyes and tell which keyboard is being used just like a person's voice.

Regarding the keyboard layouts, each one of IBM's keyboard layouts influenced a different regional layout. The IBM XT layout influenced the ISO layout, the IBM AT layout influenced the Asian layout, and the IBM AT Extended (Model M) layout influenced the ANSI layout.

[sharktastica]

IBM beamsprings have been made even in the mid-80's.

As a supporting source for this, you can check out IBM document S126-0029-0 from January 1983. Even in this 'late' stage, IBM was maintaining a parts catalogue for some beam spring keyboards (in this case, 3276/3278).

[XMIT]

.doc

Please use one of the many open file formats. PDF, text file, .odf, or just pasting inline with a "spoiler" tag come to mind.

It's plain text. I gave it a .doc extension only to pacify the uploader - it wouldn't take .adoc or .txt.

Indeed it is. This is a forum bug, we should fix this.

[Findecanor]

Cool that you know Jay Maynard (aka "The TRON guy") and chose to have him.

Thank you for taking the opportunity to set things straight about rollover and ghosting. The misconceptions about those things are a frequent annoyance on forums.

I'm not sure I myself would give an "honorable mention" to Drop (what Massdrop is called now) myself, and while I think Matias and Das Keyboard deserve to be honoured for things they've done in the past, I wouldn't recommend their keyboards these days. Matias' Alps-clone switches are known to have some issues, and Das Keyboard has been quite stagnant in recent years while competitors have stepped up.
But as long as you don't recommend any "gaming" brand, I'm fine.

[esr]

Cool that you know Jay Maynard (aka "The TRON guy") and chose to have him.

Jay and I have been friends for a very long time. I knew him before he was Tron Guy.

I'm not sure I myself would give an "honorable mention" to Drop (what Massdrop is called now) myself, and while I think Matias and Das Keyboard deserve to be honoured for things they've done in the past, I wouldn't recommend their keyboards these days. Matias' Alps-clone switches are known to have some issues, and Das Keyboard has been quite stagnant in recent years while competitors have stepped up.

I was thinking of the Drop CTRL, which by every report I have is quite good. Though I've never used one myself.

This is undoubtedly the area where my knowledge is weakest. If you had to recommend three vendors who are pushing towards better typing ergonomics, who would be?

[MMcM]

There are, I will claim, better vintage keyboards. But it isn't practical to work any of them into this, so the hyperbole is fine.

I would say "mid-80's" for the VT220. You couldn't realistically have bought one before FY84.

I hope the M122 you have has the proofreaders delete mark and not a boring word legend.

I would see if you can work in that Thomas is a chemist by day, as that makes keyboard enthusiasts seem a little less weird.

[Weezer]

IBM beamsprings have been made even in the mid-80's.

As a supporting source for this, you can check out IBM document S126-0029-0 from January 1983. Even in this 'late' stage, IBM was maintaining a parts catalogue for some beam spring keyboards (in this case, 3276/3278).

I just want to add that the displaywriter beamspring (type A) keyboard was offered all the way up until the displaywriter was pulled from the market in 1985 or so. However IBM continued to produce beamspring keyboards as warranty replacements until the late 80's and early 90's. I have a displaywriter keyboard from 1988. This keyboard was not refurbed.

[esr]

Okay, so...

These are useful comments. In rough order:

I mentioned Drop because of the CTRL, as I said to Findecanor. If you have better candidates for keyboard vendors who are pushing in the direction of better typing ergonomics, I'd be happy to hear about them.

No, the whole thing is not about buckling springs. It's intended to be about keyboards with good enough ergonomics to prevent RSI and last-forever durability. The fact that buckling springs dominate that category is a historical accident - that was the point of bringing up the OmniKey. Well, that and the opportunity for a little self-deprecating humor.

Other well-respected switches: I feel like trying to go down that road would have at least two major risks. One is that we'd end up trying to cover more detail than our time and the audience's attention will sustain. The other is that none of today's "well-respected" switches have yet demonstrated the kind of staying power I'm interested in. It could literally take another decade before that's even possible.

So I have to say "not going there", but I feel kind of bad about that. I'm open to suggestions creative enough to change my mind.

More on form factor and physical layout: I'd be open to adding a bit more about this. If you could add three sentences to the exposition, how would you do that? Note that I already give the pre-ANSI layout as a strike against the F; would it be sufficient to emphasize that more?

I'll ask Wendell whether he has an F AT to put on the specimen table. I'm afraid that's what will control whether we get into the difference between the F AT and F XT.

Terminal variants: Getting into details like conversion strategies is out of scope for this video. My Troubleshooting FAQ covers that stuff like a blanket, which is why I have Wendell nudge the reader to go look at it.

[sharktastica]

It's almost 2 am here and I can't sleep, so I decided to make some more comments. Hopefully, they're helpful (and the humour is well received). I originally had a few more comments, but acting on them would likely lengthen the script/video even more so I've chosen the most concise ones:

• During the "DISPLAY BUCKLING-SPRING ANIMATION", you mention "capacitive sensor". Obviously, this is in reference to OG capacitive buckling springs but you've only mentioned Model M at this point. Maybe, adding the phrase "as originally conceived" somewhere in that sentence will make it absolutely clear you're referring to OG buckling springs and that there's more to the story.
• I've already mentioned "Soft Touch" in my first comment. On second thought, I doubt the casual audience would have an issue with this since it's clear you're referring to rubber domes being soft. I'm not sure if you plan to have any additional comments in the video description, but it might be worth making a footnote that in IBM's terminology, "Soft Touch" was greased buckling springs and "Quiet Touch" was rubber dome.
• Around "eye-wateringly expensive to make", maybe it's worth mentioning that they're also extremely tall switches and almost everyone in the industry moved to smaller switches in the early '80s as well. For example, Alps SKCC > SKCL/SKCM and Cherry M7 > MX, etc. Although, I'm unsure if IBM's reasoning included the introduction of the relevant DIN standards (since buckling springs was released a year before DIN 66234-6 and DIN 66234-7 were drafted, but IBM may have known they're coming). In any case, an industry trend existed.
• At "don't have the ANSI layout of a modern keyboard", maybe it's worth saying "ANSI or ISO" to keep the Europeans like me happy.
• If indeed you're happening to be doing any footnotes, an interesting fact to add would be that membrane buckling springs were actually first used in late 1984 for the IBM Quietwriter 7 (source, ignore that it calls it "Selectric" as well). On that note, another interesting fact is that capacitive buckling springs were also used in 1982's IBM Electronic Typewriter Models 65, 85 and 95.
• Around "A few particularly cheaped-out Lexmark Model Ms were even made with rubber-dome switches, which is BLASPHEMY.", maybe add as a joke "Well, they at least Lexmark added drainage holes!" On that note, maybe a mention/sidenote here that you can identify these Quiet Touch Ms via their part number starting with "71G"

[esr]

I just want to add that the displaywriter beamspring (type A) keyboard was offered all the way up until the displaywriter was pulled from the market in 1985 or so. However IBM continued to produce beamspring keyboards as warranty replacements until the late 80's and early 90's. I have a displaywriter keyboard from 1988. This keyboard was not refurbed.

I have corrected appropriately.

[esr]

Maybe you could elaborate on the beamspring vs model F vs model M key feel and sound.

We'll probably get to that in the unscripted typing-demo part.

[esr]

There are, I will claim, better vintage keyboards. But it isn't practical to work any of them into this, so the hyperbole is fine.

Out of curiousity, which ones are you thinking of?

I would say "mid-80's" for the VT220. You couldn't realistically have bought one before FY84..

If I say "mid-1980s" that would suggest an interval centered on 1985, rather than ending there or before. That would be especially confusing because of the Model M's release date. So I think I have to stick with "early".

[Findecanor]

I was thinking of the Drop CTRL, which by every report I have is quite good. Though I've never used one myself.

I'm not saying it isn't a good keyboard, it's the part about "Drop" being in the same sentence as "honourable" that I think people could have objections to ... because of drama.
BTW. The keyboard had originally been designed by Input Club and sold by Drop with another colour scheme as the "K-Type". Except for the switches, it is Open Hardware, but I think only Drop has ever manufactured it.

I think recommendations (or anything that could be viewed as recommendations) should come from you personally, based on your own experiences.
Overall, I think that the world of Cherry MX-type switch keyboards is moving pretty fast these days and that options for high-quality keyboards are not scarce any more.

[depletedvespene]

…

More on form factor and physical layout: I'd be open to adding a bit more about this. If you could add three sentences to the exposition, how would you do that? Note that I already give the pre-ANSI layout as a strike against the F; would it be sufficient to emphasize that more?

Careful there. Don't confuse the Enhanced Layout, which is the form factor + physical layout introduced with the Model M keyboards, with its "ANSI" and "ISO" variants (which refer to the differing arrangements of the left Shift and Enter keys). See http://www.farah.cl/Keyboardery/Just-ho ... -Refer-to/ for more information.

…

Terminal variants: Getting into details like conversion strategies is out of scope for this video. My Troubleshooting FAQ covers that stuff like a blanket, which is why I have Wendell nudge the reader to go look at it.

Then you need to make a clear separation between "terminal" keyboards and "PC" keyboards (as the F/M122 keyboards are strictly the former).

[esr]

During the "DISPLAY BUCKLING-SPRING ANIMATION", you mention "capacitive sensor". Obviously, this is in reference to OG capacitive buckling springs but you've only mentioned Model M at this point. Maybe, adding the phrase "as originally conceived" somewhere in that sentence will make it absolutely clear you're referring to OG buckling springs and that there's more to the story.

I'm going to take the cowardly, simplifying way out and remove "capacitive".

I've already mentioned "Soft Touch" in my first comment. On second thought, I doubt the casual audience would have an issue with this since it's clear you're referring to rubber domes being soft. I'm not sure if you plan to have any additional comments in the video description, but it might be worth making a footnote that in IBM's terminology, "Soft Touch" was greased buckling springs and "Quiet Touch" was rubber dome.

I've reached the same conclusion independently. Thanks for confirming.

Around "eye-wateringly expensive to make", maybe it's worth mentioning that they're also extremely tall switches and almost everyone in the industry moved to smaller switches in the early '80s as well. For example, Alps SKCC > SKCL/SKCM and Cherry M7 > MX, etc. Although, I'm unsure if IBM's reasoning included the introduction of the relevant DIN standards (since buckling springs was released a year before DIN 66234-6 and DIN 66234-7 were drafted, but IBM may have known they're coming). In any case, an industry trend existed.

OK, that's too deep in the weeds for this video. But I'm glad you told me about it, because it's the kind of thing I want to know.

At "don't have the ANSI layout of a modern keyboard", maybe it's worth saying "ANSI or ISO" to keep the Europeans like me
happy.

Done.

If indeed you're happening to be doing any footnotes, an interesting fact to add would be that membrane buckling springs were actually first used in late 1984 for the IBM Quietwriter 7 (source, ignore that it calls it "Selectric" as well). On that note, another interesting fact is that capacitive buckling springs were also used in 1982's IBM Electronic Typewriter Models 65, 85 and 95.

That's also too deep in the weeds for this video. However, I have expanded and improved the Wikipedia article on the Model M, rather a lot recently, and I think this would make a great addition to it. This is me encouraging you to do that thing

Around "A few particularly cheaped-out Lexmark Model Ms were even made with rubber-dome switches, which is BLASPHEMY.", maybe add as a joke "Well, they at least Lexmark added drainage holes!"

That is a good idea and I will look for a way to work it in.

On that note, maybe a mention/sidenote here that you can identify these Quiet Touch Ms via their part number starting with "71G"

Weeds.

[esr]

I think recommendations (or anything that could be viewed as recommendations) should come from you personally, based on your own experiences.

This runs head-on into the problem that I have almost no personal experience of any of the modern mechanicals. The only exception is a Leopold FC900R (Cherry Browns) that I bought my wife as an experiment. She didn't like it much; she, like me, is a heavy typist, and is much happier with the Unicomp Classic I gave her when I replaced it with a New Model M. I thought it was aggressively mediocre.

In order to write that section I asked several close geek friends who are aware of buckling springs but for various reasons use other things who they thought the vendors doing the best typing ergonomics were. Those were the names that came up, and they seemed reasonable in view of things I had read while doing my research,

The high-level problem here is that I want to acknowledge that things in the gamer end of the pool aren't as bad as they used to be for people who care about typing ergonomics, and have a reasonable prospect of getting better still. If you can think of a better way to do that than to name vendors who you think are coasting on past glories, I'm open to suggestions.

And I know nothing about the drama you allude to.

[Rayndalf]

This is undoubtedly the area where my knowledge is weakest. If you had to recommend three vendors who are pushing towards better typing ergonomics, who would be?

Maltron (~1977), Kinesis (~1991, Advantage and associated boards), and the Keyboardio (2017, Model 01).
Open source projects such as the Ergodox and Atreus are also worthy of mention.

I don't remember which one, but either Maltron or Kinesis references an IBM patent for a split keyboard dating back to the 60s, so it's safe to say the ideas have been out there for a while.

Xah Lee's website has a wealth of information about the history of ergoboards.
http://xahlee.info/kbd/ergonomic_keyboards_index.html

[Rayndalf]

I was thinking of the Drop CTRL, which by every report I have is quite good. Though I've never used one myself.

I'm not saying it isn't a good keyboard, it's the part about "Drop" being in the same sentence as "honourable" that I think people could have objections to ... because of drama.
BTW. The keyboard had originally been designed by Input Club and sold by Drop with another colour scheme as the "K-Type". Except for the switches, it is Open Hardware, but I think only Drop has ever manufactured it.

I think recommendations (or anything that could be viewed as recommendations) should come from you personally, based on your own experiences.
Overall, I think that the world of Cherry MX-type switch keyboards is moving pretty fast these days and that options for high-quality keyboards are not scarce any more.

Yeah, the CTRL is just a hotswap keyboard with a rather simplistic metal enclosure. Sure it's metal but they've cut every possible corner (look at the PCB if you ever get a chance, the traces are really strange).

The Halo switches are for all intents and purposes an Cherry MX clear stem with a longer shaft so that there is no post bump/post actuation travel. This is nothing revolutionary. They hyped it up as "similar foce curve to Topre", but Topre is a collapsing dome, not something you like. They only came into their own when people shoved them into Invyr Panda housing which have a stronger contact leaf. The made the bump even larger, but I don't think Holy Pandas are any kind of ergonomic revelation.

If you're gonna shill anything you should talk about boards using optical or analog sensing methods, or just stick with Ellipse's stuff. Unless Drop will pay you for talking about their mediocre boards. The GMMK is still a better buy than any of Drops offerings in most peope's opinions.

Edit: Spend anytime talking about Drop talking about Unicomp instead. They're critically important in keeping old buckling spring boards healthy and do a lot to keep the price from exploding.

[MMcM]

There are, I will claim, better vintage keyboards. But it isn't practical to work any of them into this, so the hyperbole is fine.

Out of curiousity, which ones are you thinking of?

Since this is about typing experience and ergonomics, and since I have a decade or more of experience from back in the day, I'll go with the Micro Switch Hall Effect keyboards from Stanford and MIT and the various offshoots. As you know, even when they are the butt of jokes like in TGQ's cartoons, it's with a preface that they were in fact well designed for the systems with which they were used.

I won't make any claims about RSI, though, since one of the important hackers of that community had it so bad they had to use an amanuensis for many months.

[Weezer]

You just gotta put this video on EZ mode and change the title to:

"The greatest keyboards of all time" - Linus Tech Tips | A deep dive


Then you'll just be repeating Linus' opinion and can talk about how IBM keyboards are factually superior no problem and not worry about any of these obscure small potatoes like the "Maltron" and "das keyboard" and "topre". Redirect any triggered cherry boys to support@lttstore.com to complain about how Linus' opinion didn't mention keyboard x or switch y, because it's Linus' opinion, not yours. Your hands are tied.

[depletedvespene]

You just gotta put this video on EZ mode and change the title to:

"The greatest keyboards of all time" - Linus Tech Tips | A deep dive
…

Just don't say that "Lexmark/Unicomp keyboards are all rubber dome", like LS said, and the end result will be better than LTT's video.

[esr]

I won't make any claims about RSI, though, since one of the important hackers of that community had it so bad they had to use an amanuensis for many months.

Heh. You must mean RMS.

I've had my hands on a Space Cadet keyboard exactly once. I was visiting MIT sometime back in the 1980s, hanging with him, and he thought I'd find it interesting.

Well, I did.

The keyboard wasn't a legend yet. And, actually, now I think about it, RMS wasn't a legend yet either. That was a long time ago.

[esr]

and the Keyboardio (2017, Model 01).

When I met Jesse on his promotional tour he promised to make me one of those out of gunstock walnut. Said it would be a good excuse to play with his milling machine.

Occasionally I consider nudging him about this. Haven't done so yet.

[Rayndalf]

Then you'll just be repeating Linus' opinion and can talk about how IBM keyboards are factually superior no problem and not worry about any of these obscure small potatoes like the "Maltron" and "das keyboard" and "topre".

The first das keyboard was a custom order G80-3000 with blank keycaps. I'd argue they began as a successful marketing campaign but their boards alongside Filco and Razer were critical in the mechanical keyboard renaissance.

If the Model M is only the best keyboard if you want a standard fullsize with 2KRO. For many people gamer scum a TKL or 75% with Cherry switches and NKRO is better. Cherry switches aren't perfect, but I'd genuinely argue a hotswap board with discrete switches will have significantly greater longevity than a board that requires drilling out rivets. The model F on the other hand doesn't need rivet replacement. It's a little painful to disassemble, but that's usually done to take advantage of the fact that the F AT has the pads to support more modern layouts. And came after the 4704 keyboards. And shipped with a dumpster layout anyway.

[Findecanor]

This runs head-on into the problem that I have almost no personal experience of any of the modern mechanicals. The only exception is a Leopold FC900R (Cherry Browns) that I bought my wife as an experiment. She didn't like it much; she, like me, is a heavy typist, and is much happier with the Unicomp Classic I gave her when I replaced it with a New Model M. I thought it was aggressively mediocre.

Leopold has a reputation for good build quality, but the switches are the same Cherry MX as in every other board. Your wife is far from alone in thinking that the MX Brown are not tactile enough. Some people liken them to being "scratchy linears".

Another brand known for good build quality, and for their colourful design is Varmilo. Their keyboards don't look too "gamery", but no, they too use the same switches as every other brand.

There are not that many sharply tactile MX-style switch varieties. As you've probably surmised, that is difficult with their mechanical design. The ones with click jacket (MX Blue) or click-bar (some Kailh switches) are the ones that come closest, but the click-bar switches make a noise on the up-stroke too, which I find completely wrong.

There do exist MX-style switches that have a bump in the key travel like the Browns but where the bump is more pronounced, but unfortunately they are at the fringe and not in the mainstream.
At the risk of blaspheming, I would say that the beginning of a key stroke feels like a rubber dome, but after actuation like a linear.
The first one out was the Cherry MX Clear (which should better have been named Cherry MX "Translucent") with a quite heavy spring. Then there are clone /alike switches with lighter springs collectively known as "Ergo Clears" (including the "Halo" and "Holy Panda" switches previously mentioned in the thread). Some Ergo Clears, especially sound-dampened ones, fetch stupid high prices - per switch, available only separately, for the custom DIY crowd.

The Cherry MX Clear is my personal favourite, as I don't mind the spring being sharply heavy below the actuation point. In fact, I quite like that it cushions my key presses, thus helping keep my presses short. They take some time to get used to though, so many who have tried them discarded them too quickly.
Leopold, Ducky and Varmilo have occasionally made keyboards with them.

Drop and "Glorious PC Gaming Race" ( ) regularly offer Holy Panda and Glorious Panda respectively, (the latter being a clone of the former) for an extra cost. The bump is early in the key travel and the back is dragged out — which is the opposite characteristic to that of a buckling spring, but at least it is more tactile than an MX brown.

Yeah, the CTRL is just a hotswap keyboard with a rather simplistic metal enclosure. Sure it's metal but they've cut every possible corner (look at the PCB if you ever get a chance, the traces are really strange).

I don't think the strange-looking PCB design is a result of cutting corners, but more of the designer's (lack of) style ...
I've never heard about any problems leading from the ugly routing though.